Journal of Sociolinguistics 11/4, 2007: 478–504 Phonation type as a stylistic variable: The use of falsetto in constructing a persona1 Robert J. Podesva Georgetown University, Washington, D.C. Although the field of sociolinguistics has witnessed a growing interest in the sociophonetic aspects of segmental and intonational variation, few studies have examined variation in voice quality. This paper addresses the gap by investigating the stylistic use of falsetto phonation. Focusing on the speech of Heath, a speaker exhibiting considerable cross-situational variation, I show that when attending a barbecue with friends, Heath’s falsetto is more frequent, longer, and characterized by higher fundamental frequency (f0) levels and wider f0 ranges. Advancing recent approaches to variation which treat linguistic features as stylistic resources for constructing social meaning, I draw on an analysis of the discourse contexts in which falsetto appears to illustrate that the feature carries expressive connotations. This meaning is employed to construct a ‘diva’ persona and may also participate in building a gay identity. KEYWORDS: Phonation type, falsetto, stylistic variation, social meaning, persona, sexuality 1. INTRODUCTION In recent years, the field of sociolinguistics has witnessed a growing interest in the sociophonetic aspects of segmental variation, and to a lesser extent, intonational variation. In spite of great advances in these areas, few studies have examined variationinvoicequality.Thispaperaddressesthegapbyinvestigatingthestylistic use of the falsetto phonatory setting as well as creaky voice in the speech of Heath, a gay medical school student who exhibits a considerable phonetic range across situations. Situating Heath’s falsetto utterances in their conversational contexts, I infer the social meaning of falsetto and show that variable phonation can be used as a resource to construct personae and perhaps gay identity. Voice quality, also called phonetic setting, refers to the extragrammatical suprasegmental properties of speech resulting from the configuration of the vocal apparatus. As Foulkes (2002) points out, the term voice quality sometimes refers to supralaryngeal properties of speech like nasality, but is often used to describe only phonation type (e.g. creaky and breathy voice). Honikman (1964) and two decades later Scherer (1986) lamented the underexplored status of voice quality C The author 2007 Journal compilation C Blackwell Publishing Ltd. 2007 9600 Garsington Road, Oxford OX4 2DQ, UK, and 350 Main Street, Malden MA 02148, USA PHONATION TYPE AS A STYLISTIC VARIABLE 479 in linguistics. Scherer suggests that methodological hurdles could be responsible in part for the scarcity of research on voice quality; there is reason to believe that the extreme complexity of the relationship between the production of voice quality and its acoustic consequences (Laver 1980; Nolan 1983) has deterred sociolinguists from investigating the phenomenon more thoroughly. Another twenty years have now transpired, and the number of studies on voice quality variation remains scant (Foulkes 2002, 2005; Thomas 2002). Nonetheless, a few noteworthy studies deserve mention. Honikman (1964) was perhaps the first to explicitly comment on social axes of voice quality variation, impressionistically describing differences in oral setting between English and French. A number of other studies have followed a similar methodology,relying on trained phoneticians to describe the auditory properties of the voices under investigation. Most of these studies have aimed to characterize the voice qualities associated with particular speech communities, with Trudgill (1974) focusing on Norwich, Knowles (1978) on Scouse, Esling (1978) on Edinburgh, and Stuart-Smith (1999) on Glasgow. While distinct voice qualities appear to mark each of these communities, voice quality also distinguishes between groups within them. For example, when compared to middle-class speakers, working-class speakers in the Esling and Stuart-Smith studies made greater use of harsh voice and whispery voice, respectively. Other studies have used the matched guise technique to access listeners’ attitudes toward various voice qualities. Pittam (1987) shows that both Australian and American listeners accorded high status to English-speaking males with tense voices, consistent with Esling’s (1978) observation that high prestige speakers in Edinburgh used more creaky voice. Pittam additionally reports that the same listeners attributed higher status to female speakers with breathy voices. A final group of studies has turned to an acoustic approach for investigating voice quality. Pittam and Gallois (1986) correlate impressionistic labels for the breathy, creaky, tense, and whispery voice qualities with a variety of distinct acoustic properties. Investigating breathy voice and creaky voice among speakers of two dialects of British English, Henton and Bladon (1985, 1988) report a higher degree of breathiness among female speakers and a higher degree of creaky voice among male speakers. In addition to exhibiting sex-based differences in phonation, creaky voice distinguishes between speakers of the two dialects, with Northern speakers producing more creak than RP (Received Pronunciation) speakers. Although these studies have brought about a much greater understanding of the link between voice quality features and the social factors affecting their use, they leave two research gaps. First, while they have attended to most kinds of phonation, such as breathy,creaky,and whispery voice, they have not considered another commonly employed phonation type: falsetto, a phonatory setting in which the vocal folds are adducted tightly and stretched lengthwise, resulting in a high fundamental frequency (Esling 1984).2 Second, although these studies C The author 2007 Journal compilation C Blackwell Publishing Ltd. 2007 480 PODESVA have examined how voice quality patterns are distributed across speakers, none examine intraspeaker variation patterns. Implicit in the approach of previous studies is the belief that voice quality is a static characteristic of individuals, that different voice qualities distinguish one social group from another.3 This study seekstopartiallyfillthesegapsbyinvestigatingtheintraspeaker,stylisticvariation patterns of a speaker who employs falsetto phonation. The glottal configuration for falsetto gives rise to rapid vocal fold vibration, correlating acoustically with a high fundamental frequency (f0) level which can range from 240 Hz to 634 Hz in the speech of men (Hollien and Michel 1968; Svec, Schutte and Miller 1999). This contrasts sharply with the average modal voice f0 level for adult men of approximately 100 Hz (e.g. 106 Hz in Eady 1982). Using the high f0 levels characteristic of falsetto phonation is a socially marked behavior,atoddswithmoreculturallynormativepitchpracticesformen,andmay be involved in the performance of stereotypical gay identity. Indeed, much of the research on male pitch patterns has focused on the potential role of fundamental frequency as a marker of gay male identity. Work on the pitch properties of gay-sounding voices began with Gaudio’s (1994) experimental study on listeners’ perceptions of the speech of gay and straight men. After recording speakers in a laboratory setting, Gaudio asked listeners to identify the sexual orientation of the speakers. He found that even though listeners were adept at correctly identifying whether or not speakers were gay, neither f0 range nor f0 variability provided sufficient cues to yield a gay percept. These findings suggest that f0 is not the only phonetic resource available for performing gayness, and have inspired a number of researchers to look beyond prosody and consider the potential role of segmental variables in performances of gayness (Crist 1997; Linville 1998; Rogers, Smyth and Jacobs 2000; Podesva, Roberts and Campbell-Kibler 2002; Pierrehumbert, Bent, Munson, Bradlow and Bailey 2004; Munson, McDonald, DeBoe and White 2005; Levon 2006; Munson, Jefferson and McDonald 2006). The most comprehensive study to investigate pitch as an acoustic correlate of perceived gay male identity is the ongoing project of Smyth, Jacobs, and Rogers (2003). They recorded 25 subjects (17 gay, 8 straight) speaking three kinds of text (scientific and dramatic reading passages as well as responses to an open- endedquestiondesignedtoelicitaspontaneousspeechsample).Forty-sixlisteners were then asked to categorize speakers as gay/straight or masculine/feminine. Smyth, Jacobs and Rogers found that although listeners appeared to associate pitch with masculinity/femininity, there was no correlation between pitch and sexual orientation ratings. Thus, even though in many English-speaking communities the lay person commonly identifies high pitch as a property of gay-sounding voices, the link between the phonetic feature and the identity category has not been established empirically in any community, not even in the speech of a single individual. The observed tendency for gay-sounding men to resist using f0 stylistically in these studies, I argue, stems less from characteristics of the speakers themselves C The author 2007 Journal compilation C Blackwell Publishing Ltd. 2007 PHONATION TYPE AS A STYLISTIC VARIABLE 481 than from the constraints imposed on them by the research paradigm used to analyze their speech. Speaking with high f0 levels is not simply a way to sound gay, but a way to sound flamboyant, and reading passages are unlikely